chain renaming: update also MODRES

refactor and unify renaming functions: change_ccd_code() was renamed to
rename_residues() and moved from polyheur.hpp to modify.hpp

include/gemmi/modify.hpp

@@ -111,6 +111,11 @@ inline void assign_serial_numbers(Structure& st, bool numbered_ter=false) {
 }
 
 
+/// Helper function for processing (usually: changing) names and numbers
+/// in AtomAddress instances in metadata:
+/// Connection, CisPep, Helix, Sheet::Strand.
+/// Other fields are not updated here, in particular: ModRes, Entity::DbRef,
+/// Entity::full_sequence, TlsGroup::Selection.
 template<typename Func>
 void process_addresses(Structure& st, Func func) {
   for (Connection& con : st.connections) {
@@ -134,24 +139,53 @@ void process_addresses(Structure& st, Func func) {
     }
 }
 
-
 inline void rename_chain(Structure& st, const std::string& old_name,
                                         const std::string& new_name) {
-  process_addresses(st, [&](AtomAddress& aa) {
-      if (aa.chain_name == old_name)
-        aa.chain_name = new_name;
-  });
+  auto update = [&](std::string& name) {
+    if (name == old_name)
+      name = new_name;
+  };
+  process_addresses(st, [&](AtomAddress& aa) { update(aa.chain_name); });
+  for (ModRes& modres : st.mod_residues)
+    update(modres.chain_name);
   for (RefinementInfo& ri : st.meta.refinement)
     for (TlsGroup& tls : ri.tls_groups)
       for (TlsGroup::Selection& sel : tls.selections)
-        if (sel.chain == old_name)
-          sel.chain = new_name;
+        update(sel.chain);
+  for (Model& model : st.models)
+    for (Chain& chain : model.chains)
+      update(chain.name);
+}
+
+inline void rename_residues(Structure& st, const std::string& old_name,
+                                           const std::string& new_name) {
+  auto update = [&](ResidueId& rid) {
+    if (rid.name == old_name)
+      rid.name = new_name;
+  };
+  process_addresses(st, [&](AtomAddress& aa) { update(aa.res_id); });
+  for (ModRes& modres : st.mod_residues)
+    update(modres.res_id);
+  for (Entity& ent : st.entities)
+    for (std::string& mon_ids : ent.full_sequence)
+      for (size_t start = 0;;) {
+        size_t end = mon_ids.find(',', start);
+        if (mon_ids.compare(start, end-start, old_name) == 0) {
+          mon_ids.replace(start, end-start, new_name);
+          if (end != std::string::npos)
+            end = start + new_name.size();
+        }
+        if (end == std::string::npos)
+          break;
+        start = end + 1;
+      }
   for (Model& model : st.models)
     for (Chain& chain : model.chains)
-      if (chain.name == old_name)
-        chain.name = new_name;
+      for (Residue& res : chain.residues)
+        update(res);
 }
 
+
 inline void rename_atom_names(Structure& st, const std::string& res_name,
                               const std::map<std::string, std::string>& old_new) {
   auto update = [&old_new](std::string& name) {

include/gemmi/polyheur.hpp

@@ -191,16 +191,8 @@ inline void trim_to_alanine(Chain& chain) {
 
 // Functions for switching between long (>3 chars) residue names (CCD codes)
 // and shortened ones that are compatible with the PDB format.
-GEMMI_DLL
-void change_ccd_code(Structure& st, const std::string& old, const std::string& new_);
-
 GEMMI_DLL void shorten_ccd_codes(Structure& st);
-
-inline void restore_full_ccd_codes(Structure& st) {
-  for (const auto& item : st.shortened_ccd_codes)
-    change_ccd_code(st, item.second, item.first);
-  st.shortened_ccd_codes.clear();
-}
+GEMMI_DLL void restore_full_ccd_codes(Structure& st);
 
 } // namespace gemmi
 #endif

prog/convert.cpp

@@ -265,7 +265,7 @@ void convert(gemmi::Structure& st,
     std::string new_name(sep+1);
     if (options[Verbose])
       std::cerr << "Renaming " << old_name << " to " << new_name << std::endl;
-    gemmi::change_ccd_code(st, old_name, new_name);
+    gemmi::rename_residues(st, old_name, new_name);
   }
 
   if (options[AssignRecords])

src/polyheur.cpp

@@ -2,6 +2,7 @@
 
 #include <gemmi/polyheur.hpp>
 #include <gemmi/resinfo.hpp>   // for find_tabulated_residue
+#include <gemmi/modify.hpp>   // for rename_residues
 
 namespace gemmi {
 
@@ -328,51 +329,6 @@ bool trim_to_alanine(Residue& res) {
   return true;
 }
 
-void change_ccd_code(Structure& st, const std::string& old, const std::string& new_) {
-  auto process = [&](ResidueId& rid) {
-    if (rid.name == old)
-      rid.name = new_;
-  };
-  for (Model& model : st.models)
-    for (Chain& chain : model.chains)
-      for (Residue& res : chain.residues)
-        process(res);
-  for (Entity& ent : st.entities)
-    for (std::string& mon_ids : ent.full_sequence)
-      for (size_t start = 0;;) {
-        size_t end = mon_ids.find(',', start);
-        if (mon_ids.compare(start, end-start, old) == 0) {
-          mon_ids.replace(start, end-start, new_);
-          if (end != std::string::npos)
-            end = start + new_.size();
-        }
-        if (end == std::string::npos)
-          break;
-        start = end + 1;
-      }
-  for (Connection& conn : st.connections) {
-    process(conn.partner1.res_id);
-    process(conn.partner2.res_id);
-  }
-  for (CisPep& cispep : st.cispeps) {
-    process(cispep.partner_c.res_id);
-    process(cispep.partner_n.res_id);
-  }
-  for (ModRes& modres : st.mod_residues)
-    process(modres.res_id);
-  for (Helix& helix : st.helices) {
-    process(helix.start.res_id);
-    process(helix.end.res_id);
-  }
-  for (Sheet& sheet : st.sheets)
-    for (Sheet::Strand& strand : sheet.strands) {
-      process(strand.start.res_id);
-      process(strand.end.res_id);
-      process(strand.hbond_atom2.res_id);
-      process(strand.hbond_atom1.res_id);
-    }
-}
-
 template <size_t I, typename T1, typename T2>
 bool in_vector_at(T1& x, std::vector<T2>& v) {
   for (const auto& el : v)
@@ -410,7 +366,7 @@ void shorten_ccd_codes(Structure& st) {
     if (!in_vector_at<1>(short_code, st.shortened_ccd_codes))
       old_new.second = short_code;
   }
-  // pick a new residue name and call change_ccd_code()
+  // pick a new residue name and call rename_residues()
   int i = -1;
   for (auto& old_new : st.shortened_ccd_codes) {
     // If ~DE was not unique, use ~00, ~01, ...
@@ -421,8 +377,15 @@ void shorten_ccd_codes(Structure& st) {
       if (!in_vector_at<1>(short_code, st.shortened_ccd_codes))
         old_new.second = short_code;
     }
-    change_ccd_code(st, old_new.first, old_new.second);
+    rename_residues(st, old_new.first, old_new.second);
   }
 }
 
+void restore_full_ccd_codes(Structure& st) {
+  for (const auto& item : st.shortened_ccd_codes)
+    rename_residues(st, item.second, item.first);
+  st.shortened_ccd_codes.clear();
+}
+
+
 } // namespace gemmi